Hi all, I'll try list my problem in the following below. Data: from RT-PCR experiment (log(2^-delta delta t) data) Experiment: (A1,A2,A3,B1,B2.B3) samples belonging to 2 groups (A,B) and want to find the differential expressed genes Statistical test : t-test or use limma between 2 groups. It works fine. Software : R [Packages: genefilter,limma] Problem: We know that A3 is slightly different from the rest of the samples by viewing the boxplot so I want to find the differentially expressed genes of A3 against the group B (B1,B2 and B3)?.This means 1 sample against a group of 3. As it is I don't have enough samples and using a t-test using sample A3 and mean of (B1,B2,B3) seems statistically incorrect for this purpose, I tried using LIMMA package thinking it should produce sensible results. Would this be correct or should I be using any other tests. Any help rendered is appreciated. cheers, John.

Hi John, That's not a huge amount of info to go on. Is there any a priori reason for A3 to be different? Or have you just deduced this from the boxplots? Why do you believe A3 over A2/A2 to be the "correct result"? I don't understand why you wish to compare A3 on it's own to B1-3. By comparing 1 sample against 3 using t.test, as I understand it, you will be conducting a one-sample t-test, which means you hypothesise that the A3 is the population mean (which it most likely isn't, since it's from 1 sample, and especially because it appears to be an outlying sample). I suspect you might not get a reproducible result in this manner. Also, Perhaps if the delta Cqs are all higher/lower in A3 compared to A2 and A1, something went wrong in normalisation. Perhaps check the expression of your reference genes. Best wishes, Jim On 27 February 2012 15:07, John Alexander <fbsja@leeds.ac.uk> wrote: > > Hi all, > I'll try list my problem in the following below. > Data: from RT-PCR experiment (log(2^-delta delta t) data) > Experiment: (A1,A2,A3,B1,B2.B3) samples belonging to 2 groups (A,B) and > want to find the differential expressed genes > Statistical test : t-test or use limma between 2 groups. It works fine. > Software : R [Packages: genefilter,limma] > > Problem: We know that A3 is slightly different from the rest of the > samples by viewing the boxplot so I want to find the differentially > expressed genes of A3 against the group B (B1,B2 and B3) .This means 1 > sample against a group of 3. As it is I don't have enough samples and using > a t-test using sample A3 and mean of (B1,B2,B3) seems statistically > incorrect for this purpose, I tried using LIMMA package thinking it should > produce sensible results. Would this be correct or should I be using any > other tests. > Any help rendered is appreciated. > cheers, > John. > _______________________________________________ > Bioconductor mailing list > Bioconductor@r-project.org > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: > http://news.gmane.org/gmane.science.biology.informatics.conductor > [[alternative HTML version deleted]]